Chunking Strategy for Retrieval Augmented Generation in Regulation Documents
Amar Fadillah, Nuke Athahirah, Kuan Ting Lai · 2024
Large language model (LLM) have been proven to be capable of performing various text processing tasks. LLMs have the potential to be used as a tool for analyzing regulation documents. To enhance the ability of LLMs to analyze regulation documents, we propose a regulation document chunking method that improves the accuracy of the analysis results. Our proposed method produces better accuracy than the sequential chunking method.